Genes Predicting Survival of Chordoma Patients

Predicting overall survival in chordoma patients using machine learning models: a web-app application

OBJECTIVE

The goal of this study was to evaluate the efficacy of machine learning (ML) techniques in predicting survival for chordoma patients in comparison with the standard Cox proportional hazards (CoxPH) model.

METHODS

Using a Surveillance, Epidemiology, and End Results database of consecutive newly diagnosed chordoma cases between January 2000 and December 2018, we created and validated three ML survival models as well as a traditional CoxPH model in this population-based cohort study. Randomly, the dataset was divided into training and validation datasets. Tuning hyperparameters on the training dataset involved a 1000-iteration random search with fivefold cross-validation. Concordance index (C-index), Brier score, and integrated Brier score were used to evaluate the performance of the model. The receiver operating characteristic (ROC) curves, calibration curves, and area under the ROC curves (AUC) were used to assess the reliability of the models by predicting 5- and 10-year survival probabilities.

RESULTS

A total of 724 chordoma patients were divided into training (n = 508) and validation (n = 216) cohorts. Cox regression identified nine significant prognostic factors (p < 0.05). ML models showed superior performance over CoxPH model, with DeepSurv having the highest C-index (0.795) and the best discrimination for 5- and 10-year survival (AUC 0.84 and 0.88). Calibration curves revealed strong correlation between DeepSurv predictions and actual survival. Risk stratification by DeepSurv model effectively discriminated high- and low-risk groups (p < 0.01). The optimized DeepSurv model was implemented into a web application for clinical use that can be found at https://hust-chengp-ml-chordoma-app-19rjyr.streamlitapp.com/ .

CONCLUSION

ML algorithms based on time-to-event results are effective in chordoma prediction, with DeepSurv having the best discrimination performance and calibration.

SMARCB1 Loss in Poorly Differentiated Chordomas Drives Tumor Progression

Poorly differentiated (PD) chordoma, a rare, aggressive tumor originating from notochordal tissue, shows loss of SMARCB1 expression, a core component of the Switch/Sucrose Non-Fermentable (SWI/SNF) chromatin remodeling complexes. To determine the impact of SMARCB1 re-expression on cell growth and gene expression, two SMARCB1-negative PD chordoma cell lines with an inducible SMARCB1 expression system were generated. After 72 hours of induction of SMARCB1, both SMARCB1-negative PD chordoma cell lines continued to proliferate. This result contrasted with those observed with SMARCB1-negative rhabdoid cell lines in which SMARCB1 re-expression caused the rapid inhibition of growth. We found that the lack of growth inhibition may arise from the loss of CDKN2A (p16INK4A) expression in PD chordoma cell lines. RNA-sequencing of cell lines after SMARCB1 re-expression showed a down-regulation for rRNA and RNA processing as well as metabolic processing and increased expression of genes involved in cell adhesion, cell migration, and development. Taken together, these data establish that SMARCB1 re-expression in PD chordomas alters the repertoire of SWI/SNF complexes, perhaps restoring those associated with cellular differentiation. These novel findings support a model in which SMARCB1 inactivation blocks the conversion of growth-promoting SWI/SNF complexes to differentiation-inducing ones, and they implicate SMARCB1 loss as a late event in tumorigenic progression. Importantly, the absence of growth inhibition after SMARCB1 restoration creates a unique opportunity to identify therapeutic vulnerabilities.

LIM and SH3 protein 1 (LASP1) differentiates malignant chordomas from less malignant chondrosarcomas

PURPOSE

Chordomas are malignant tumors that develop along the neuraxis between skull-base and sacrum. Chondrosarcomas show similarities with chordomas, yet show less malignant behavior. LIM and SH3 protein 1 (LASP1) is a cytoskeletal protein known to promote the malignant behavior of tumors. LASP1 was previously identified as a possibly overexpressed protein in a chordoma proteomics experiment. In this study we compare LASP1 expression in chordoma and chondrosarcoma tissue.

METHODS

Biopsies of primary tumors were collected from surgically treated chordoma (n = 6) and chondrosarcoma (n = 6) patients, flash-frozen upon collection and collectively analyzed for LASP1 RNA (real-time PCR) and protein expression (western blotting). Additionally, tissue micro array (TMA)-based immunohistochemistry was applied to an archive of 31 chordoma and 1 chondrosarcoma specimen.

RESULTS

In chordoma samples, LASP1 mRNA was detected in 4/6 cases and a strong 36 kDa immunoreactive protein band was observed in 4/5 cases. In contrast, 0/6 chondrosarcoma samples showed detectable levels of LASP1 mRNA and only a weak 36 kDa band was observed in 4/5 cases. Immunohistochemical analysis showed LASP1 expression in all chordoma samples, whereas chondrosarcoma specimen did not show immunoreactivity.

CONCLUSION

LASP1 is strongly expressed in the majority of chordoma cases and shows low expression in chondrosarcoma tissue. Since LASP1 is known to function as oncogene and regulate cell proliferation in other tumor types, this study implicates a role for LASP1 in chordoma biology. Further studies are warranted to improve understanding of LASP1's expression and functioning within chordoma, both in vitro and in vivo.